• Spectroscopy and Spectral Analysis
  • Vol. 41, Issue 2, 572 (2021)
You-fu LIU*, De-qin XIAO, and Chun-tao WANG
Author Affiliations
  • College of Mathematics and Informatics, South China Agricultural University/Guangdong Province Agricultural Data Engineering Research Center, Guangzhou 510642, China
  • show less
    DOI: 10.3964/j.issn.1000-0593(2021)02-0572-07 Cite this Article
    You-fu LIU, De-qin XIAO, Chun-tao WANG. Fertilized Eggs’ Air-Cell Change Monitoring Algorithm Based on Thermal-Image[J]. Spectroscopy and Spectral Analysis, 2021, 41(2): 572 Copy Citation Text show less
    Collecting thermal images
    Fig. 1. Collecting thermal images
    Schematic diagram of Faster-RCNN
    Fig. 2. Schematic diagram of Faster-RCNN
    Flow chart of segment
    Fig. 3. Flow chart of segment
    Multi-channel grayscale(a): Default grayscale; (b): R channel; (c): G channel; (b): B channel
    Fig. 4. Multi-channel grayscale
    (a): Default grayscale; (b): R channel; (c): G channel; (b): B channel
    Otsu segment(a): Thermal image of breeding eggs; (b): Default grayscale; (c): Otsu algorithm
    Fig. 5. Otsu segment
    (a): Thermal image of breeding eggs; (b): Default grayscale; (c): Otsu algorithm
    Structure of BP neural network
    Fig. 6. Structure of BP neural network
    Trend of total loss function
    Fig. 7. Trend of total loss function
    Result of eggs detection
    Fig. 8. Result of eggs detection
    Some segment of egg’s thermal images(a): Thermal images; (b): Otsu algorithm; (c): BP neural network
    Fig. 9. Some segment of egg’s thermal images
    (a): Thermal images; (b): Otsu algorithm; (c): BP neural network
    Segment’s images of egg and air cell
    Fig. 10. Segment’s images of egg and air cell
    Change of air cell size
    Fig. 11. Change of air cell size
    Correlation between the artificial measured value and thermal-image measured value
    Fig. 12. Correlation between the artificial measured value and thermal-image measured value
    仪器参数参数值
    热像仪测量范围-20~+120 ℃
    测温精度±0.05 ℃
    图像分辨率320×240 pixels
    波长范围7.5~13 μm
    孵化箱控温精度±0.1 ℃
    电热功率300 W
    容蛋量(鸡蛋)1 232枚
    Table 1. Instrument parameter value
    初始学习速率动量系数平均精度均值/%
    0.1-0
    0.01-96.57
    0.001-97.63
    0.000 1-68.85
    0.599.65
    0.0010.999.74
    0.9999.85
    Table 2. Learning rate and momentum coefficient
    隐藏层结构F1度量/%
    1 00081.45
    1 000, 1 00085.76
    1 000, 1 000, 1 00090.81
    1 000, 2 000, 1 00090.94
    1 000, 3 000, 1 00091.19
    3 000, 3 000, 3 00089.91
    Table 3. Hidden layers’ optimization of BP neural network
    初始学习速率最大迭代次数F1度量/%
    0.169.76
    0.0180.01
    0.0011 00088.89
    0.000 190.38
    0.000 0188.17
    0.000 110085.64
    30088.58
    50090.17
    80090.12
    Table 4. Hyperparameters’ optimization of BP neural network
    图像状况Otsu算法效果/%BP神经网络效果/%
    只存在一个蛋68.8687.17
    有两个蛋存在61.6486.94
    总体65.2587.02
    Table 5. Comparison of algorithms for segment
    You-fu LIU, De-qin XIAO, Chun-tao WANG. Fertilized Eggs’ Air-Cell Change Monitoring Algorithm Based on Thermal-Image[J]. Spectroscopy and Spectral Analysis, 2021, 41(2): 572
    Download Citation